How Yanxuan Reduces Stockouts: Layered Inventory Management Design

1. Basic Methods of E‑commerce Inventory Management

E‑commerce has moved from chaotic growth to refined operations, and its information systems have matured from simple inventory‑procurement‑sales (IPS) to full‑featured ERP, then to specialized inventory and order management systems.

1.1 Core Problems Solved by Inventory Management

With the rise of new‑retail, many e‑commerce companies now operate both online and offline, facing dispersed traffic and higher customer timeliness demands. Consequently, they adopt omni‑channel, multi‑warehouse distribution strategies, and a good inventory system must address:

Warehouse operation requirements : Real‑time recording of inbound, outbound, and inventory‑checking actions, possibly tracking SKU, location, batch, and expiration.

Decision support for order fulfillment : In a multi‑warehouse scenario, the chosen warehouse directly impacts delivery speed and cost; fulfillment decisions rely on whether inventory is sufficient.

Flexible support for complex marketing sales models : The system must handle single‑ or multi‑channel, nationwide or regional, regular, promotional, or offline sales while preventing overselling.

1.2 E‑commerce Inventory Layering System

Different business scenarios (warehouse, order fulfillment, front‑end sales) have distinct inventory data needs, making a single data model insufficient. The common solution is a layered inventory design :

Warehouse‑Level Inventory

Manages data directly from WMS or store‑management systems, reflecting the physical quantity of goods in each warehouse or store. Dimensions may include owner, location, batch, etc., and any operation (picking, shipping, shelving, counting) affects this layer.

Physical‑Level Inventory

Aggregates warehouse data into a “dispatch inventory” used for order scheduling. It does not track detailed locations but may be managed by SKU, region, or total quantity. In Yanxuan, this layer also tracks position (in‑stock, in‑transit) and batch information, enabling calculations of expiration and age.

Sales‑Level Inventory

Exposed to customers, representing the sellable quantity. It derives from physical inventory but is not identical; for example, if a product has 2 units in Warehouse A and 2 in Warehouse B and a user purchases 1 unit, the sales‑level inventory drops to 3 while the physical inventory remains 4 until the order is allocated.

Imagine a product with 2 sellable units in each of two warehouses (total 4). After a user buys 1, the sales‑level available stock becomes 3, while the physical stock stays at 4 until allocation.

Sales‑level inventory typically tracks SKU‑level quantities, sometimes further split into activity or channel inventories.

2. Yanxuan Inventory Center Design Practice

Yanxuan follows the layered design principle but adapts it to its specific business characteristics.

2.1 Yanxuan’s E‑commerce Business Features

Omni‑channel, multi‑scenario sales : Yanxuan sells via its own app, enterprise purchase platform, and third‑party stores (Taobao, Tmall, JD), plus offline stores. Most channels share inventory, and the app hosts dozens of promotional activities, each with unique stock requirements.

Multi‑warehouse, nationwide shipping : Warehouses in North, South, East, and Southwest China can ship to any region, so there is no regional inventory concept; all stock is shared nationwide.

Outsourced warehousing : Yanxuan uses third‑party WMS (e.g., SF, JD) and does not own direct warehouse data.

2.2 Yanxuan Inventory Layer Design

Because Yanxuan does not control warehouse‑level data, its inventory center focuses on two layers: sales‑level and physical‑level .

Sales‑level : Managed only at SKU granularity, reflecting the nationwide shared stock.

Physical‑level : Enriched with location, batch, expiration, and age dimensions to support detailed supply‑chain decisions.

The two layers are linked by inventory formulas so that any change in physical stock automatically updates the sales view.

2.3 Yanxuan Lockable Stock Design

Beyond the standard layered model, Yanxuan introduces a flexible lockable stock pool to handle various special‑use cases such as promotional activities, B2B contracts, or region‑specific replenishment.

All shared inventory forms a large pool; within it, independent lockable pools (each identified by a unique lockkey) can be created. Business units request a lockable pool by providing lockkey + sku + quantity. Once allocated, they use only the lockkey to consume stock, ensuring isolation from other operations.

Lockable stock pools allow activities to reserve dedicated inventory without affecting the general sellable stock.

This mechanism supports numerous scenarios:

Multiple concurrent promotions each with its own reserved stock.

Third‑party platform rules (e.g., Tmall during Double‑11) requiring guaranteed inventory.

Channel‑specific pre‑stock for offline stores or B2B contracts.

Special logistics cases such as cost‑optimized shipments from designated warehouses.

While the design simplifies rapid response to business needs, challenges remain, including weak governance of lockable pools, inconsistent lockkey standards, and higher learning costs for downstream systems.

Conclusion

The core of e‑commerce inventory management lies in a layered data model—warehouse, physical, and sales layers—tailored to business requirements. After establishing this foundation, the real value comes from supply‑chain optimization: minimizing cost (low stock turnover) while ensuring fast, region‑balanced fulfillment, which involves sophisticated procurement planning and inter‑warehouse balancing.